1 Method to measure the size of a radiographic field larger than

2 a detector by imaging fluorescence x-rays with a slit camera

3
4
5
6

7 ABSTRACT

8 Purpose: Clinical x-ray imaging devices contain a collimator system that defines a rectangular irradiation field on the
9 detector plane. The size and position of the x-ray field, and its congruence with the corresponding light field, must be
10 regularly tested for quality control. We propose a new method to estimate how far the x-ray field extends beyond the
11 detector which does not require the use of external digital detectors or film strips.

12 Methods: A metallic foil is inserted perpendicularly between the source and the detector. A slit camera, a linear
13 extension of a pinhole camera, is used to project onto the detector the fluorescence x-rays emitted by the irradiated
14 foil. The location where the fluorescence signal inside the camera vanishes is used to extrapolate the location of the
15 field boundary. Monte Carlo simulations were performed to determine the optimal composition and thickness of the
16 foil. A prototype camera with a 1-mm-wide slit was built and tested in a clinical mammography and digital breast
17 tomosynthesis (DBT) system.

18 Results: The simulations estimated that a foil made of 25 µm of Molybdenum provided the maximum signal inside
19 the camera for a 39 kVp beam. The boundary of the x-ray fields in mammography and DBT views were experimentally
20 measured. With the camera along the chest wall side the measured field in multiple DBT views had a variability of
21 only 0.4±0.1 mm compared to mammography. With the camera along the right side an expected change in the field
22 position of ±6 mm was observed.
23 Conclusion: The introduced technique provides a practical alternative method to detect the boundary of an x-ray field.
24 The method can be combined with other testing methods to assess the congruence of the x-ray and light fields, and to
25 determine if the x-ray field extends beyond the detector more than permitted.

26 Keywords: x-ray field, collimator, slit camera, fluorescence, radiography, mammography, quality control
27

1
28 1. INTRODUCTION

29 One of the essential components of a clinical x-ray imaging system is a beam-limiting device that defines a rectangular
30 field-of-view (FOV) on the detector plane. The beam-limiting device is typically constructed with two pairs of
31 orthogonal collimator plates. Some systems also have a mirror below the focal spot that reflects a light beam through
32 the beam-limiting device, providing a visible projection of the FOV that simplifies patient positioning before
33 irradiation. To maximize the recorded anatomical information, the collimated x-ray field is commonly slightly larger
34 than the rectangular image receptor. However, radiation emitted outside the detector delivers unnecessary radiation
35 dose to the patient and produces unwanted scatter. A misalignment between the x-ray and light fields can result in
36 faulty images that need to be retaken. For these reasons, methods to evaluate the congruence between the x-ray field,
37 the light field, and the active area of the image receptor have been created.

38 In the particular case of mammography systems, regulatory agencies mandate strict limits on the collimation accuracy,
39 and prescribe corresponding quality control activities that need to be performed in devices used in clinical practice. In
40 the United States, the Code of Federal Regulations 1 mandates that film-based mammographic beam-limiting devices
41 “shall be provided with a means to limit the useful beam such that the x-ray field at the plane of the image receptor
42 does not extend beyond any edge of the image receptor by more than 2% of the source-to-image detector distance
43 (SID).” The Mammography Quality Standards Act Regulations (MQSA) 2 regulations also require an “X-ray
44 field/light field/image receptor/compression paddle alignment” assessment as part of the annual quality control tests
45 for all screen-film systems (MQSA Sec. 900.12 (e)(5)(vii)). This regulation mandates that the mammographic beam
46 must extend beyond the chest wall side of the image receptor but not by more than 2% of the SID. In addition, if the
47 system has a light field it shall be aligned with the x-ray field so that the sum of the misalignments in length and width
48 does not exceed 2% of the SID. For full-field digital mammography and digital breast tomosynthesis (DBT) devices,
49 similar limits in the fields congruence are required by the applicable manufacturer quality control protocols or by the
50 quality assurance program recommended by the ACR Digital Mammography Quality Control Manual 3. The ACR
51 quality control manual leaves to the discretion of the medical physicist the choice of the most appropriate tool to test
52 the collimation congruence, and lists as possible options: film, computed radiography plate, self-developing film, or
53 electronic radiation rulers. The European guidelines for quality assurance in breast cancer screening and diagnosis 4
54 also recommend the use of film, computed radiography plates, or self-developing film for the “Alignment of X-ray
55 field/image receptor” test (sections 2a.2.1.1.3 and 2b.2.1.1.3), and specify that the x-ray field shall not extend more
56 than 5 mm outside the detector.

57 Current test methods to assess the congruence between the x-ray field and the detector assume that the source is static
58 and perpendicular to the detector plane. These two assumptions do not hold in DBT projections. In this modality the
59 x-ray field may extend substantially beyond the detector at large acquisition angles if the collimator aperture is fixed.
60 There are not established methods to measure the x-ray field in individual DBT views. In one of the few publications
61 that have addressed this issue, Popova et al. 5 used radiosensitive film and an electronic radiation ruler with customized
62 acquisition software to measure the FOV in DBT views. The aim of Popova’s work was to assess the performance of
63 a dynamic collimation system implemented in GE Healthcare’s SenoClaire DBT system *, which moves the collimator
64 blades during the DBT sweep to keep the FOV close to the detector boundaries.

65 In this article we propose a new method to estimate how far the x-ray field extends beyond the detector active area.
66 The method can be used in radiography and in DBT system. The main advantage is the use of the standard image
67 detector instead of an external sensor or film. An example use in a clinical mammography and DBT system is
68 presented.

69

70

*
The mention of commercial products herein is not to be construed as either an actual or implied endorsement of
such products by the Department of Health and Human Services.

2
 71 2. MATERIALS AND METHODS

72 2.A. SLIT CAMERA MEASURING SYSTEM

73 We propose to use the built-in detector in a clinical imaging system to measure the distance to the outer limit of the
74 x-ray field beyond the detector by transforming the transmission image acquisition into an emission image acquisition.
75 The emission signal consists of secondary radiation produced by an object blocking the primary x-ray beam at a known
76 height above the detector. A slit camera, a linear extension of a pinhole camera, is used to project the secondary
77 radiation onto the detector, enabling the visualization of the field boundary in the acquired image.

78 When a thin, flat foil of material is positioned above and parallel to the detector plane, some of the primary x-rays
79 coming from the x-ray source will be absorbed by photoelectric interactions with the foil atoms. After photoelectric
80 absorption, the atoms are ionized in a high energy state and quickly absorb an electron and emit the excess energy in
81 the form of Auger electrons or one or more fluorescence (or characteristic) x-rays with specific energies. These
82 secondary particles are emitted in random directions. Some other primary x-rays will not interact with the foil, while
83 others will change direction due to Compton or Rayleigh scattering interactions. A slit camera located between the
84 foil and the detector is able to project the secondary radiation emitted in the foil onto the detector. The purpose of the
85 projection process is to allow imaging the boundary of the beam located beyond the detector inside the detector.

86 Figure 1 provides a simple diagram of the operation of the slit camera with the fluorescent foil in a mammography
87 system. The diagram shows how to measure the size of the FOV at the right side of the detector. Analogous
88 measurements with the foil and camera at different sides of the detector can measure the FOV in the left, chest wall
89 and nipple sides. In ideal conditions, the top plate of the camera should completely block the primary beam, and the
90 image captured by the detector should show an uninterrupted signal from fluorescence radiation in the projection of
91 the part of the foil inside the FOV, and no signal beyond the FOV. The position of the abrupt transition between the
92 two regions is determined by the position of the boundary of the FOV beyond the detector.

93 A slit camera, which captures much more light than a pinhole camera, can be used in this application because the edge
94 we are imaging is a straight line. It is worth noting that in the quality control of x-ray imaging systems, slit cameras
95 have been used extensively to measure the size of the focal spot 6. We use the same projection principle, but imaging
96 secondary radiation instead of the primary beam directly. Slit cameras, pinholes and other general coded apertures
97 have been used in other radiological imaging applications such as SPECT 7, prompt gamma imaging in proton therapy 8,
98 or small-angle x-ray scattering measurements.

99 Figure 2 presents pictures of a prototype slit camera system installed on a Hologic Selenia Dimensions mammography
100 and DBT system (Hologic, Inc., Marlborough, MA). The camera was built using 6.35-mm-thick brass plates †. Brass
101 alloy 360, with a nominal composition of 61.5% Copper (Cu), 35.5% Zinc (Zn), 3% Lead (Pb), and density 8.5 g/cm3,
102 was chosen because it has a large attenuation coefficient for mammographic beams, and has the practical advantages
103 of being low-cost and easily machinable. The top of the camera was a 10×10 cm brass plate, suspended above the
104 breast support by two 10×5 cm side brass plates and 1 mm Aluminum (Al) spacer plates. In this mammography system,
105 the breast support is 2.5 mm thick and located 25.0 mm above the detector plane (the anti-scatter grid can be inserted
106 between the support and the detector, but it was not used in our experiments). The front of the camera was a 10×5 cm
107 brass plate positioned approximately 0.5 mm ahead of the top plate. Thanks to the Al spacers, the lower part of the
108 top plate was 1 mm above the upper part of the front plate. Therefore, the slit aperture size was 1 mm high, 0.5 mm
109 wide and 100 mm long. In practice, the effective size of the aperture seen by the x-rays depends on the incidence
110 angle. The back of the camera was covered with a large Pb plate to minimize the area of the detector irradiated by the
111 primary beam. A 15×10 cm metal foil was framed to remain flat, and suspended on 38.5-mm-tall plastic blocks above
112 the breast support, with the edge of the light beam delimiting the irradiation field aligned with the center of the frame.

†
The brass plates of the appropriate size were sourced from https://www.cut2sizemetals.com. The metal foils were
purchased from Alfa Aesar, part of Thermo Fisher Scientific (Tewksbury, MA).

3
113 The slit was also aligned parallel to the light field edge, and positioned approximately 30 mm away from the edge
114 towards the center of the detector.

115 The distance from the side of the detector (i.e., the first or last column or row of pixels, which we define as the origin
116 of coordinates) to the end of the x-ray FOV beyond the detector (xFOV) can be calculated using Eq. 1. This equation
117 (which was deduced using the similarity of triangles in the diagram in Fig. 1) uses the position of the edge inside the
118 slit camera measured in the image profiles (xedge) and the known positions of the foil, slit, and x-ray source.
𝒚𝒚𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇 𝒚𝒚𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇
𝒙𝒙𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆 +� �⋅�𝒙𝒙𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔 −𝒙𝒙𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆 �−� �⋅𝒙𝒙𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔
𝒚𝒚𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔 𝒚𝒚𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔
119 𝒙𝒙𝑭𝑭𝑭𝑭𝑭𝑭 = 𝒚𝒚𝒇𝒇𝒇𝒇𝒇𝒇𝒇𝒇 (1)
𝟏𝟏−� �
𝒚𝒚𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔𝒔

120 The horizontal position of the source focal spot (xsource) is near half the detector width when measuring the lateral side
121 of the detector, and near 0 when measuring the chest wall side. The exact position in three dimensions of the focal
122 spot in our experimental setting was measured with a geometric calibration phantom composed of ball bearings in
123 known positions in two planes separated by 75.5 mm, as described by Li, Zhang and Liu. 9

124 Determining the exact position of the edge inside the camera (xedge) is challenging because the edge is expected to be
125 blurred by the finite aperture of the slit, x-ray transmission through the slit walls, unwanted signal from scatter
126 generated in the breast support, and inherent detector resolution limits. In addition, the boundary of the FOV is not
127 sharp (penumbra) due to the finite size of the focal spot, transmission through the collimator edge and presence of off-
128 focal spot radiation. For a reliable and repeatable measure, the edge location was extrapolated from a linear
129 interpolation fitted to the points in the steep intensity fall leading to the edge.

130

131 2.B. FLUORESCENT FOIL OPTIMIZATION

132 There are five main requirements on the fluorescent foil material used in the proposed slit camera: a high x-ray
133 attenuation to allow the use of a thin foil that emits radiation at a well-defined plane; a K-edge energy low enough that
134 a substantial fraction of the x-rays emitted by a diagnostic imaging source can produce fluorescence; a large
135 fluorescence yield after photoelectric effect; a Kalpha fluorescence energy sufficiently high to pass through the detector
136 cover and provide a measurable signal; and the commercial availability of stable, solid foils of the material. In an
137 alternative embodiment of the proposed method, a material that produces a large amount of scatter could be utilized,
138 but this option was not studied in this work.

139 For the application of the proposed method in mammography, which is our first objective, Molybdenum (Mo)
140 appeared as a natural option for the foil material, because this metal is used as target material and filter in many
141 commercial mammography systems. Since the Mo K-edge is used to limit the maximum energy in clinical beams, its
142 Kalpha emission, with an energy slightly below than the K-edge, should have an energy within the optimal detection
143 range of mammography detectors, while being energetic enough to pass through the breast support and detector cover.
144 Other elements with an atomic number right below Mo, such as Zirconium (Zr), are of interest because their lower K-
145 edge energy enables the generation of fluorescence from x-ray generated in a source using a Mo target and Mo filter.
146 Table 1 provides the fluorescence parameters for Mo and other elements of interest as provided in the NIST reference
147 databases 10 11

148 Table 1 shows that Cu, a material commonly used in x-ray imaging experiments, has a low fluorescence yield and a
149 characteristic energy that is too low to be reliably detected below the breast support plate of a mammography machine.
150 Zr and Mo have high yields and an optimal fluorescence energy for a mammography detector. In addition, thin foils
151 of these materials are chemically stable and readily available for purchase from chemical product suppliers. The mean
152 free path for photoelectric interactions in brass is 22.3 µm for Mo Kalpha fluorescence, 17.0 µm for Zr, and 42.8 µm
153 for Ag. Therefore, the 6.35-mm-thick brass plates used in the prototype camera were about 150 times thicker than the
154 mean free path of the most energetic x-rays in the Ag-filtered mammography beam. In principle, this attenuation was

4
155 more than necessary to block the primary beam, but was adequate for an initial prototype in which the total weight
156 and ease of use are not relevant factors.

157 The optimal thickness of the foil material depends on the balance between two competing requirements: thick foils
158 maximize primary radiation absorption and the consequent secondary radiation generation, while thin foils maximize
159 secondary radiation transmission towards the slit camera. The optimal foil thickness and composition was estimated
160 using Monte Carlo simulations. The radiation transport simulation code PENELOPE 2018 12 with the modular
161 penEasy 13 main program was used to simulate the primary x-rays, scatter, and fluorescence emission in a digital twin
162 of the proposed device. PENELOPE uses state-of-the-art models of fluorescence emission, including partial cross
163 sections for the K shell and L, M, and N subshells of all neutral atoms 14. The geometry of the imaging system
164 (schematically shown in Fig. 1) can be easily modelled by intersecting volumes delimited by quadric surfaces 15, which
165 is the standard geometry model used in PENELOPE. In our setting, the geometry requires only intersecting multiple
166 planes, and a cylinder for the Pb collimator edge to produce a realistic penumbra at the limit of the field.

167 The simulated x-ray beam was emitted from a cubic focal spot with 350 µm side, and reproduced the energy spectrum
168 of a clinical source with a Tungsten (W) target, 39 kVp voltage, and 60 µm Silver (Ag) filtration (mean energy of 22.5
169 keV). The simulated photon and electron tracks were modelled down to 10 keV. For the foil thickness optimization,
170 the scored quantity of interest was the total energy arriving at a 5×5 cm region of interest at the center of the camera.
171 This quantity estimates the magnitude of the signal detected inside the camera from fluorescence x-rays generated in
172 the foil. The simulation was repeated for a range of thicknesses, and for both Zr and Mo foils.

173 The penEasy imaging detector tally was also used to simulate mammographic images of the slit camera device. The
174 simulations allowed us to study the transmission of the primary beam through the top plate and the corners of the slit;
175 study the contribution of fluorescence, Compton, and Rayleigh radiation generated in the foil and in the breast support
176 to the signal detected inside the camera; and estimate the sensitivity of the device to measure small changes in the
177 field size.

178

179 2.C. VALIDATION WITH EXTERNAL SENSORS

180 The proposed methodology to measure the size of a radiographic field needs to be compared to established, commonly
181 used methods to determine if the new process provides acceptable results. As an initial step in the validation process,
182 we measured the field size in our mammography system using two commercial products: a RTI Nova optically
183 fluorescent ruler set (RTI Group AB. & Inc., Mölndal, Sweden), and Gafchromic XR self-developing film (Ashland
184 Global Holdings Inc., Wilmington, Delaware). The extern sensors were positioned across the chest side of the detector,
185 and the position of the edge of the light field was marked with a metal indicator on the ruler and with a pen line on the
186 film. A mammographic exposure was made with the same parameters used with the slit assembly experiments. The
187 room was kept in the dark, and a cellphone camera recorded the optical fluorescence emitted by the rulers.

188

189 3. RESULTS

190 The results of the PENELOPE simulations estimating the signal detected inside the slit camera for a range of Mo and
191 Zr foil thicknesses are provided in Fig. 3. The results show that the maximum signal was obtained with 25 µm of Mo.
192 The maximum signal with Zr, only 6% smaller than the Mo maximum, was obtained with a thickness of 50 µm. The
193 optimization process was not further refined for other intermediate thicknesses near the peaks because there are not
194 commercially available foils at arbitrary thicknesses.

195 Figure 3 also presents experimental measurements with the prototype device in the Hologic Selenia Dimensions
196 mammography machine using multiple layers of 25 µm foils. To compare with the simulations, the experimental
197 results were normalized to match the simulated signal at 25 µm for the corresponding material.

5
198

(a) 39 kVp spectrum (b) 100 kVp spectrum

199 Fig. 3. Optimization of the foil material and thickness using PENELOPE simulations, and corresponding experimental
200 results, for 39 kVp (a) and 100 kVp spectra (b). The reported signal was the average pixel value inside the slit camera.
201 The simulated results were normalized by the largest reported signal. The two sets of experimental results were
202 separately normalized to match the simulated signal with 25 µm of material.

203

204

205 Experimental projection images of the prototype slit camera acquired with the clinical system are shown in Fig. 4.
206 Sub-figures (b) and (c) show the signal acquired with the x-ray field collimated to 18×24 cm and 24×29 cm (full field),
207 respectively. The two field sizes were defined using the source built-in Pb collimator. The former field was smaller
208 than the detector and its limits can be seen in the radiographic image (except in the chest wall side); the latter full field
209 covered the entire detector. Sub-figures (d) and (e) show the signal in the DBT projections at 0 degrees (equivalent to
210 a mammogram) and at the maximum angulation of 7.5°. ‡ For the FOV smaller than the detector shown in sub-figure
211 (a), the distance from the top of the image to the x-ray field boundary, C, was 21.1 mm, while the distance to the
212 boundary of the light field marked by a radiopaque needle, D, was 19.0 mm.

213 The acquired images had 3328×4096, 70 µm pixels in mammography, and 1664×2048, 140 µm in DBT (2×2 binning).
214 The manual imaging settings used for the mammography acquisitions were 39 kVp and 450 mAs, which were the
215 maximum values allowed by the system. A 50-µm-thick Ag filter was used to produce a harder beam (i.e., containing
216 more high energy x-rays above the Mo K-edge) than the default Rhodium filter. For the DBT projections, 39 kVp and
217 125 mAs were used. The system automatically used a 700-µm-thick Al filter in DBT mode, and therefore the beam
218 contained even more high energy x-rays than with the Ag filter because Al does not have an absorption edge in the
219 energy range of interest. The built-in antiscatter-grid would absorb the fluorescence radiation and interfere with the
220 slit camera operation, and therefore it was retracted for all acquisitions.

221 Figure 5 summarizes the information in the ROI images in Fig. 4 in the form of line intensity profiles. The profiles
222 were generated by integrating the width of the ROIs, and then convolving with a 5-pixel-wide mean filter for noise
223 reduction. The integration assumes that the slit was accurately aligned with the side of the detector to avoid blurring
224 the fluorescence edge. Each profile was normalized to its minimum intensity inside the camera, to compensate for

‡
The individual DBT projection images were obtained from the secondary capture image file thanks to a proprietary
conversion software provided by Hologic, Inc.

6
225 differences in exposure and incidence angle in each projection. The line profiles with the slit camera aligned with the
226 chest wall side of the detector (not shown in Fig. 4) are also presented. In the chest wall configuration, the plane of
227 rotation of the source in DBT was parallel to the slit and, consequently, the field size is expected to remain equal to
228 that in the mammogram at all angles.

229 To determine the position of the fluorescence signal edge inside the slit camera, linear fits were performed to the linear
230 section of the profiles leading to the edges in Fig. 5. The fittings are shown in Fig. 6. The edge position (xedge) was
231 found by solving the fitting equations for y=1. Table 2 presents the measured edge positions and the corresponding
232 FOV positions calculated with Eq. 1. All distances are reported with respect to the nearest side of the detector, with
233 negative distances representing a position outside of the detector and positive distances inside the detector. The
234 following geometric parameters were used in the equation: yfoil=63.5 mm; yslit=31.35 mm; xslit=50.0 mm (right side),
235 33.0 mm (chest wall side); ysource=661.2 mm; xsource= 133.7 mm (right side), 23.6 mm (chest wall side). These
236 parameters were measured on the physical prototype with a caliper or post-processing the radiographic image.

237

(b)

(c)

(d)
(a)
238 Fig. 7. Measuring the position of the x-ray field beyond the chest side of the mammography detector using an RTI
239 Nova optically fluorescent ruler and a Gafchromic XR film strip: (a) picture of the external sensors on the breast
240 support; (b) video frame showing the ruler fluorescence during the irradiation; (c) flatbed scan of the self-developed
241 film after irradiation (a white horizontal line marks the position of the light field edge); (d) x-ray projection image
242 showing the radiopaque ruler markings and the light field marker slit.

243

244 Finally, Fig. 7 shows the measurement of the end of the x-ray field in the chest side of the mammography detector
245 using two external sensors. The edge of the irradiation field and the markers of the light field edge were clearly visible
246 in a frame of the video recording of the optically fluorescent ruler (Fig. 7 (b)), and also in the self-developed film (Fig.
247 7 (c)). The distance between the light and x-ray fields was measured to be 11.2 mm in the fluorescent ruler, and 7.9
248 mm in the film. These measurements were taken on top of the breast support, and need to be scaled to the detector
249 plane 25.0 mm below. The measurements had an uncertainty of the order of 1 mm due to the penumbra effect that,

7
250 expectedly, prevented having sharp edges in both the light and x-ray fields. The metallic marker indicating the position
251 of the light field edge on the ruler was visible in the x-ray projection image (Fig. 7 (d)). The distance between this
252 light field edge marker and the end of the detector was 1.5 mm, as measured in the image knowing the pixel size of
253 70 µm. Projecting the ruler and film measurements to the detector plane and subtracting the measured distance to the
254 edge of the detector measured on the radiograph, we estimated that the x-ray field extended 10.1 mm beyond the
255 detector in the fluorescent ruler measurement, and 6.7 mm in the film measurement. These distances can be compared
256 to the reported distance of 7.68 in the first row of Table 2 for the slit camera experiment.

257

258 4. DISCUSSION

259 The presented methodology successfully measured the distance to the x-ray field boundary in a clinical system. The
260 measured fluorescence signal below the camera was extremely small. The pixel values output by the mammography
261 system ranged between 50 (detector offset) and 16383 (14 bit) analog-to-digital units (ADU). The pixels in the part
262 of the camera irradiated by fluorescence had values of the order of 150 ADU, while the pixels under the camera
263 without fluorescence signal had values of the order of 100 ADU, due to unwanted scatter generated outside the camera
264 and transmission through the camera top plate. Therefore, the fluorescence signal had a very low magnitude of the
265 order of 50 ADU above the background. Fortunately, it was possible to reliably measured this low signal thanks to the
266 integration of many pixels into a line profile, as seen in Fig. 5. A longer slit could be used to integrate more pixels in
267 the line profiles. Multiple acquisitions or larger mAs could also be used to reduce the signal to noise ratio.

268 Despite we did not yet validate our field measurements with an external sensor, the values reported in the table are
269 within the expected range. The measurement in the chest wall side of the detector for the DBT and mammography
270 projections (24×29 cm filed size) had an excellent agreement, with an average difference between the DBT views and
271 mammography of only 0.4±0.1 mm. In the right side, the agreement between the mammogram and the DBT view at
272 0‫ ﹾ‬was 0.29 mm. These results suggest that the measurement has good repeatability and potentially a sub-millimetric
273 precision. The distance measured for the small mammography field of 18×24 cm was 20.82 mm, which is in good
274 agreement with the value of 21.1 mm measured directly on the projection image.

275 An interesting artifact can be seen in the right side DBT profiles in Fig. 5. A sudden increase in the profile intensity
276 can be clearly seen around 70 mm. The intensity increase happens at the projection of the location where the Mo foil
277 finishes, indicating that there is some transmission of the beam through the top of the camera in the absence of foil.
278 This feature is not visible in the mammography profiles because the Ag filter in the source effectively eliminates the
279 higher energy rays that can penetrate the thick brass plate, unlike the Al filter used in DBT. In the same figure, a large
280 intensity is seen around 50 mm for the DBT projection at +7.5°. This is caused by the primary beam entering directly
281 thorough the slit at that angulation.

282 The distance to the x-ray field edge beyond the chest side of the detector measured by the two external sensors had a
283 discrepancy of 1 mm for the film and 2.5 mm for the fluorescent ruler when compared to the new method. An error
284 of the order of magnitude of a millimeter is comparable to the order of the uncertainty in the position of the edges
285 caused by the penumbra (which also might explain most of the discrepancy between the ruler and the film). This
286 provides an initial experimental evidence that the new method is comparable to the two established methods using
287 external sensors. More validation activities (and done in the four sides of the detector to understand the impact of the
288 heel effect, for example) are needed to completely characterize the new method. We expect that upgraded versions of
289 the prototype slit camera will improve the accuracy and repeatability of the measurement.

290

291 5. CONCLUSIONS
292 This article introduced a new approach to measure the congruence between the x-ray field, the light field and the
293 image detector using x-ray fluorescence and a slit camera. The main advantage of the method is the use of the built-
294 in detector in the system to measure x-ray fields that extend beyond its physical limits, avoiding the need to use

8
295 external radiation sensors for initial beam alignment and quality control. An optimization process based on Monte
296 Carlo simulations determined that a 25-µm-thick Molybdenum foil maximized the signal detected inside the slit
297 camera (for a 39 kVp spectrum with W target and Ag filtration). A prototype of the described slit camera system was
298 built and used to measure the x-ray field size in a combined mammography and DBT acquisition with a clinical
299 imaging system. The camera was sensitive enough to detect the field size in each individual DBT projection. The
300 measurements have a good repeatability, with only 0.4±0.1 mm variation in the measured field size in the chest wall
301 side of the detector for the DBT views compared to the reference mammography acquisition.

302
303 ACKNOWLEDGMENTS

304

305

306 CONFLICTS OF INTEREST

307 The author declares no conflict of interests.

308
309 REFERENCES

1
Code of Federal Regulations. 2019. 21 CFR §1020.31(f)(3)
(https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?FR=1020.31)
2
Mammography Quality Standards Act (MQSA) . 1999. Part 900--Mammography, Subpart B--Quality Standards
and Certification (https://www.fda.gov/radiation-emitting-products/regulations-mqsa/mammography-quality-
standards-act-regulations)
3
MQSA Alternative Standard #24: Approval of an Alternative Standard for Using the Quality Assurance Program
Recommended by the ACR Digital Mammography Quality Control Manual for Full-Field Digital Mammography
Systems and Supplement for Digital Breast Tomosynthesis Mammography Systems. 2018
(https://www.fda.gov/radiation-emitting-products/regulations-mqsa/24-approval-alternative-standard-using-quality-
assurance-program-recommended-acr-digital-mammography)
4
European guidelines for quality assurance in breast cancer screening and diagnosis (4th edition). 2013 (doi:
10.2772/13196)
5
Popova Y, Hersemeule G, Klausz R, Souchay H. Description and benefits of dynamic collimation in digital breast
tomosynthesis. Radiat. Prot. Dosimetry. 2015;165:321-324
6
Everson J, Gray J. Focal-spot measurement: comparison of slit, pinhole, and star resolution pattern techniques.
Radiology. 1987:165;261-264
7
Barrett H, Furenlid L, Freed M, Hesterman J, Kupinski M, Clarkson E, Whitaker M. Adaptive SPECT. IEEE
Trans. Med. Imag. 2008:27;775-788
8
Smeets J, Roellinghoff F, Prieels D, Stichelbaut F, Benilov A, Busca P, Fiorini C, Peloso R, Basilavecchia M,
Frizzi T, Dehaes JC, Dubus A. Prompt gamma imaging with a slit camera for real-time range control in proton
therapy. Phys. Med. Biol. 2012:57;3371-3405
9
Li X, Zhang D, Liu B. A generic geometric calibration method for tomographic imaging systems with flat-panel
detectors-A detailed implementation guide. Med. Phys. 2010;37:3844-3854
10
Deslattes RD, Kessler EG, Indelicato P, de Billy L, Lindroth E, Anton J, Coursey JS, Schwab DJ, Chang J,
Sukumar R, Olsen K, Dragoset RA. X-Ray Transition Energies Database. NIST Standard Reference Database 128.
2005 (https://dx.doi.org/10.18434/T4859Z)
11
Hubbell JH, Trehan PN, Singh N, Chand B, Mehta D, Garg ML, Garg RR, Singh S, Puri S. A Review,
Bibliography, and Tabulation of K, L, and Higher Atomic Shell X‐Ray Fluorescence Yields. J Phys Chem Ref Data.
1994;23:339-364
12
Salvat F, penelope-2018: A code System for Monte Carlo Simulation of Electron and Photon Transport (OECD
Nuclear Energy Agency, document NEA/MBDAV/R(2019)1, Boulogne-Billancourt, France). 2019

9
13
Sempau J, Badal A, Brualla L. A PENELOPE-based system for the automated Monte Carlo simulation of clinacs
and voxelized geometries--application to far-from-axis fields. Med. Phys. 2011;38:5887-5895
14
Sabbatucci L, Salvat F. Theory and calculation of the atomic photoeffect. Radiat. Phys. Chem. 2016;121:122-140
15
Almansa J, Salvat-Pujol F, Diaz-Londono G, Carnicer A, Lallena AM, Salvat F, PENGEOM -- A general-purpose
geometry package for Monte Carlo simulation of radiation transport in complex material structures. Comput. Phys.
Commun. 2016;199:102-113

10